Methodology.

Source integrity

The World English Bible (eBible.org edition “engwebp”) is downloaded once, pinned by SHA-256 and kept with the site. Production never fetches it. On every build the text is regenerated from that pinned archive and must come out byte-for-byte identical.

Each of the 31,103 verses is compared with eBible’s separate verse-by-verse edition. They agree except where that edition keeps a Song of Solomon speaker label (“Beloved”, “Friends”) inside a verse; we keep those labels as headings. Five verses are numbered but footnote-only in the WEB (for example Acts 8:37); they are shown that way.

Ingestion

The text is read from USFM, the format the translators publish. Only structure is processed: paragraphs, poetry lines, psalm titles, words of Jesus, “Selah”, footnotes and the WEB’s own cross references. Strong’s number tags are removed from around words. Words, spelling, punctuation and divine names are never changed, normalised or “corrected”.

References and IDs

Books use the standard three-letter USFM codes (GEN … REV) and the English versification of the WEB. Pages have one address each: /john/3 for a chapter, /verse/john/3/16 for a verse. The reference parser accepts common forms — “Jn 3 16”, “1 John 2:1”, “Ps 23” — and, when an abbreviation could mean more than one book (“Jo”, “Ph”, “Jud”), asks instead of guessing.

Search ranking

References are recognised first. Word searches use a local MiniSearch index of the whole WEB: every word must match (the last word may be a prefix; longer words tolerate a typo), and if nothing matches all words, results containing some of them are shown and labelled. Verses containing your exact phrase come first, then by relevance. The index runs in your browser.

Topic and promise curation

Topics, promises, people, places and timeline entries are chosen by hand. Every reference is validated automatically: it must exist, and promises must sit inside the wider passage they link to. Each promise says who it was first spoken to. Descriptions are short and factual; there is no generated commentary, devotional or interpretation, and no AI is used to select or explain Scripture.

Daily selection

Today’s Scripture rotates through seven curated pools (morning, hope, wisdom, peace, prayer, gratitude, strength). Each pool is shuffled into a fixed order that depends only on the date, so a passage doesn’t return until its whole pool has been used, and everyone sees the same reading on the same local day. A psalm, a teaching of Jesus and a promise are chosen the same way.

Chronology

All dates are approximate. Each era says how settled its dates are: “no datable anchor” (Creation), “debated” (the patriarchs, the exodus, the judges — where an early and a late view are both widely held), or “approximate” (from the united kingdom on, where outside history anchors the dates within years or decades). We show one common reckoning and say so; we don’t present disputed chronology as fact.

Map uncertainty

Places are marked as “site known”, “traditional site” or “location uncertain”. Uncertain places (Mount Sinai, for example) show the traditional location with a “?” and a note naming the alternatives. Coordinates are approximate and for orientation only.

Bible Brain capability detection

Which languages have text, audio, verse timing or video is read from the Bible Brain catalogue on the server, and then shown only for Bibles and recordings on our reviewed allowlist. If something isn’t on the list, it isn’t shown — even if it exists. If Bible Brain is unavailable, audio and other languages say so and English reading carries on unaffected.

Licence review

Before any Bible Brain Bible or recording appears, its copyright notice is read, its attribution recorded, and its use checked against the Bible Brain terms: free to you, notices kept, terms linked, no offline copies, no copying or re-sharing of its text, and audio streamed from Bible Brain itself rather than redistributed. Verse cards and copy/share actions use only the public-domain WEB.